Clojurians Log v2

I got the sense that that was the intention from the screencast you did, but I didn't appreaciate it till I tried it myself. Can you say more the purpose of start-biff? Maybe where is it used and why? https://github.com/jacobobryant/biff/blob/cd775ae4d31206f9aa9f2b8af5a30c6e3f7a0a4c/src/biff/system.clj#L203

Sure. The lifecycle of a typical biff process includes roughly the following: 1. setup (read config file, start nrepl server, ...) 2. start a crux node 3. start websockets 4. set up http routes 5. write static resources 6. start the web server start-biff handles steps 2-5. If you wanted to run multiple apps from the same process, you could call start-biff again and have a separate crux node etc. for the other app. That's why start-biff doesn't include 1 or 6: those two steps should only happen once per process, not once per app. When the process starts, biff.core/-main will handle 1 and 6 for you, and it'll search for namespaces with ^:biff metadata. So you create a component/plugin in a namespace with ^:biff , and that component should include a call to start-biff. e.g. https://github.com/jacobobryant/biff/blob/master/example/src/example/core.clj#L30 I've since decided that was a mistake, by the way. at some point I'll change start-biff so that it includes 1 and 6. instead of starting the process via biff.core/-main, you'll use yourapp.core/-main and have it do something like this:

(ns yourapp.core
  ...)

(def system (atom nil))

(defn -main []
  (let [config ...]
    (reset! system (start-biff config))))

(just put up an issue to track that: https://github.com/jacobobryant/biff/issues/47)